ARIMA模型与残差自回归模型在手足口病发病预测中的应用

目的 探讨求和自回归滑动平均混合模型（autoregressive integrated moving average model，ARIMA）和残差自回归模型在我国手足口病月发病率预测中的应用，并对它们的预测效果进行比较。方法 收集2008年1月～2014年12月我国手足口病月发病率资料，用SPSS 13.0和EViews 8.0分别拟合ARIMA模型和残差自回归模型，并用2014年7月～12月的数据评价模型的预测效果。结果 ARIMA模型拟合及预测的平均相对误差（average relative error，MRE），均方误差（mean square predict error，MSE），均方根误差（root mean squared predict error，RMSE）和平均绝对误差（mean absolute error，MAE）分别为14.006，4.689，2.165，0.147； 13.565，4.416，2.101，0.133。残差自回归模型拟合及预测的MRE，MSE，RMSE和MAE分别为16.793，7.247，2.692，0.171，16.206，6.639，2.577，0.164。结论 ARIMA模型拟合及预测效果优于残差自回归模型。

Application of ARIMA model and auto-regressive model in prediction on incidence of hand-foot-mouth disease

Objective To explore the application of autoregressive integrated moving average(ARIMA) model and auto-regressive model in prediction on incidence of hand,foot and mouth disease in China and compare the predicated effect among them. Methods The data of monthly incidence of hand-foot-mouth disease from January in 2008 to December in 2014 in China was collected, SPSS 13.0 and EViews 8.0 were used to fit ARIMA model and auto-regressive model respectively, at the same time, the monthly data in July to December 2014 was used to evaluate the effect of prediction. Results The average relative error(MRE), mean square predict error(MSE), root mean squared predict error(RMSE) and mean absolute error(MAE) fitted and predicated by ARIMA model were 14.006,4.689,2.165,0.147 and 13.565,4.416,2.101,0.133 respectively. The MRE, MSE, RMSE and MAE fitted and predicated by auto-regressive model were 16.793,7.247,2.692,0.171 and 16.206,6.639,2.577,0.164 respectively. Conclusions According to the model fitness and prediction accuracy, ARIMA model is superior to the auto-regressive model with a good practical value.